The ACDC project group consists of 12 students from the University of Oldenburg and is part of the German Aerospace Centre's Institute of Systems Engineering for Future Mobility (DLR-SE). With the support of the DLR, a perception component is being developed that supplies the ADORe control system with realistic data from the CARLA simulation environment.
The aim of the project is to realise a perception component that transfers data from a simulated environment to the ADORe control system in a defined form according to the requirements of the interfaces. The sensor data is to be recorded and processed realistically. The use of sensors and the processing of the data without knowledge of ground truth data from the simulation are thus attempted to be ensured. The ADORe control system should receive this processed data correctly and be able to derive action strategies from it, with which ADORe should take over the control of a vehicle in the CARLA simulation environment using this data.
In order to understand the application domain, seminar papers were written on relevant topics. Based on the understanding gained from the seminar papers, an architecture was developed. This architecture is based on the interfaces provided by ADORe in order to ensure the best possible compatibility. Therefore, the existing architecture of ADORe is analysed in detail and components for the requirements of the ADORe interface are created. Various components are combined to form a complete perception and thus form the data source for the ADORe control system. The input data of the perception components will be realistic sensors, with which realistic data will be processed and transmitted to the ADORe control system. The challenges are the fusion of sensor data and the different image processing methods so that sensor variants can also be processed.
To start the implementation, a suitable development environment is created for the project. This development environment will be used to prepare the data for ADORe using computer vision, artificial intelligence and other known methods. This project report describes the project status at the time of the second review on 7 November 2023. The first review took place on 19 July 2023.